Papers by Amir Saffari
End-to-End Entity Resolution and Question Answering Using Differentiable Knowledge Graphs (2021.emnlp-main)
Copied to clipboard
| Challenge: | End-to-end (E2E) trained models for question answering over knowledge graphs (KGQA) are effective, but training a weakly supervised dataset is difficult. |
| Approach: | They extend the boundaries of E2E learning for KGQA to include the training of an ER component. |
| Outcome: | The proposed model is fully differentiable thanks to a recent method for building differentiably KGs. |
Have Your Text and Use It Too! End-to-End Neural Data-to-Text Generation with Semantic Fidelity (2020.coling-main)
Copied to clipboard
| Challenge: | End-to-end neural data-totext generation has faced challenges generalizing to new domains and generating semantically consistent text. |
| Approach: | They propose a neural data-to-text generation system that makes minimal assumptions about the data representation and target domain. |
| Outcome: | The proposed system achieves state of the art results on four major D2T datasets with better semantic fidelity than the state-of-the-art methods. |
What do Models Learn from Question Answering Datasets? (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing models have outperformed humans on question answering datasets, but they have yet to outperform humans on the task of question answering itself. |
| Approach: | They evaluate BERT-based question answering models on their generalizability to out-of-domain examples, responses to missing or incorrect data, and ability to handle question variations. |
| Outcome: | The proposed models outperform human baselines on the widely-used SQuAD 1.1 and SQu AD 2.0 datasets. |
CLASP: Few-Shot Cross-Lingual Data Augmentation for Semantic Parsing (2022.aacl-short)
Copied to clipboard
| Challenge: | Large Language Models excel at a low-resource level given limited data, but are unsuitable for runtime systems which require low latency. |
| Approach: | They propose a method to augment training data for a model 40x smaller (500M parameters) they use Alexa to generate synthetic data from Alexa 20B to augment the training set . |
| Outcome: | The proposed method improves low-resource SP on two datasets in low-source settings. |
mReFinED: An Efficient End-to-End Multilingual Entity Linking System (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing work assumed that entity mentions were given and skipped the entity mention detection step due to a lack of high-quality multilingual training corpora. |
| Approach: | They propose a bootstrapping mention detection framework that enhances the quality of training corpora. |
| Outcome: | The proposed framework outperforms existing work in the end-to-end MEL task while being 44 times faster. |
Expanding End-to-End Question Answering on Differentiable Knowledge Graphs with Intersection (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing models that handle single-entity questions have focused on relation following . introducing intersection improves performance on multiple-entities questions by over 14% . |
| Approach: | They propose a model that explicitly handles multiple-entity questions by implementing an intersection operation. |
| Outcome: | The proposed model improves on multiple-entity questions by over 14% on two datasets . it also improves performance on questions with multiple entities by 19% . |
Mintaka: A Complex, Natural, and Multilingual Dataset for End-to-End Question Answering (2022.coling-1)
Copied to clipboard
| Challenge: | Existing question answering models can achieve high performance on simple questions that require a single fact lookup. |
| Approach: | They introduce a multilingual question-answering dataset called Mintaka . it includes 8 types of complex questions, including superlative, intersection, and multi-hop questions . they run baselines over Mintak, which achieves 38% hits@1 in English . |
| Outcome: | The proposed model achieves 38% hits@1 in English and 31% hits@1, multilingually. |